Before they started using an AI receptionist, customers in a RingCentral study estimated they were missing 21% of their incoming calls. Later, the figure dropped to 2%.

That finding didn't come from a dashboard. It came from customers themselves, in conversations led by an AI interviewer by G2.

Kevin Tsukii, Director of Content Marketing at RingCentral, had two weeks to gather customer proof for an investor meeting. He wanted customers to explain, in their own words, what RingCentral AI Receptionist (AIR) was doing for their businesses.

So he turned to AI Custom Research. Within about a week of launch, its AI voice interviewer had completed 28 customer conversations. By Kevin’s estimate, interviews like these take up to two hours each with prep, so the AI took more than a full workweek of interviewing off the team’s plate

Since 1999, RingCentral has earned the trust of 600,000 organizations worldwide. RingCentral’s AI-powered communications connect businesses and their customers to get more done. Launched in early 2025, AI Receptionist (AIR) became the fastest adopted product in company history. When the quarterly investor report called for proof of the value users were seeing, the team had only two weeks to find it.

From 28 conversations to investor-ready proof

The study gave RingCentral the customer validation its investor meeting needed, inside the deadline.

Beyond the headline numbers:

• Investor-ready proof: Kevin Tsukii shared AI Custom Research's summary reports with RingCentral's investor relations (IR) and PR teams. When a customer stood out, he asked for more of that conversation to pull into the investor deck.
• Routing leads usage: Getting callers to the right person or department is AIR’s most-used feature, and customers rate it highly.
• Fewer missed calls: By respondents' own estimates, missed calls fell from 21% before AIR to 2% after.
• A roadmap for product: Feedback went to RingCentral's product managers as a roadmap of opportunities.
• Campaigns for social: Kevin's team worked with social on customer highlights and campaigns built from the report.
  

Customer proof on a two-week clock

For Kevin Tsukii, customer voice is how RingCentral validates its products. AI especially has to earn its place, with customer stories that show efficiency, convenience, and real dollars saved.

For the investor meeting, that meant customers speaking to specific product proof points. Reviews capture what customers choose to talk about. RingCentral needed answers to the questions it chose.

The other routes to that kind of proof take time. RingCentral's customer marketing team, a group of three, runs a multi-month process for each customer story, from securing the customer to producing the final piece.

Surveys could collect numbers, but not stories. A past customer survey took about 14 months to reach roughly 1,000 responses, and most of its questions were multiple choice. Investors needed AIR customers explaining, in their own words, what the product had changed for them. So there was a need for both quantitative and qualitative data. Before Kevin joined, the content team kept a customer survey open for about 14 months to collect roughly 1,000 responses. Neither route was feasible within a two-week window.

“We had two weeks to gather customer proof. With AI Custom Research, we were able to partner with G2 to send incentives and work with our lifecycle marketing team to get 28 interviews in just a week.”

Kevin Tsukii
Director of Content Marketing, RingCentral

Handing the interviews to an AI

RingCentral secured the responses through four moves. It took the right partner, the right customers, the right questions, and a fast way to interpret and present the answers to investors.

An AI interviewer for an AI product

Kevin first came across AI Custom Research through RingCentral's AEO efforts, during a separate, larger research study on agentic voice AI.

He already knew G2 as one of the most-cited domains in AI search and fundamentally understood its role in research aimed to building credibility in a new area. It was AI interviewers that he was less sure about.

Then he saw the fit. An AI voice interviewer asking customers about voice AI would let respondents experience the technology while they talked about it. So he tried a demo.

“At first, I was a bit skeptical of having an AI interviewer conduct interviews. Once I tried a demo myself, I was pleasantly surprised by how natural the process felt.”

Kevin Tsukii
Director of Content Marketing, RingCentral

The right customers, and the right sender

RingCentral's lifecycle marketing team identified every active AIR account and its admins, then shared the list with G2 to target the most relevant people. G2 sent the invitations and handled participant incentives.

Two email touches did the job: an invitation, then a nudge to customers who had opened it but not yet responded. Kevin noticed the sender made a difference: an invitation from G2 carried more credibility than one from RingCentral.

“G2 is one of the most-cited domains when it comes to AI search. So when you're trying to build credibility in a new area, it really makes sense to leverage a highly cited domain that's very credible and authoritative.”

Kevin Tsukii
Director of Content Marketing, RingCentral

Questions built around the proof

Custom research meant RingCentral could shape the questions. The interviews centered on the product proof points the team wanted customers to speak to, asking what they liked, what was working, and where AIR could improve.

The setup moved fast. The team went from kickoff to a final survey design in about four days, then took the AI interview themselves to refine it before putting the study into the field. About a week later, the interviews were complete.

Verbatim voices, sorted into themes

RingCentral's past research had mostly run on traditional text-based surveys, where customers picked from preset answers. AI Custom Research flipped that, letting customers answer in their own words instead of choosing one of five fixed options.

AI Custom Research then sorted those answers into themes, structured the data, and surfaced standout insights, while keeping the full verbatim responses available to read. For the researcher in Kevin, that meant scale and depth at once.

It also made the data quick to reshape. When Kevin asked the AI Custom Research team to cut it by positive and negative feedback, it came back organized for two audiences: investors and the product team.

Key takeaways for B2B marketers

• Design questions around the proof you need: G2’s Research Agents let you ask about the exact proof points your audience cares about, instead of waiting for customers to bring them up.
• Recruit with your own account data: A clean list of active accounts and admins helped G2 reach the right customers quickly.
• Let a credible third party make the ask: In Kevin's experience, outreach from G2 carried more credibility than an invitation from RingCentral alone.
• Ask for critical feedback, too: Splitting results into positive and negative gave investors proof and gave the product team a roadmap.
• Count the hours you get back: Kevin estimates 10 customer interviews, including prep, can take up to 20 hours — half of someone's workweek. AI Custom Research runs those conversations for you.

What's next: Customer validation at scale

For RingCentral, one investor deadline opened a new way to listen. The team plans to keep using AI Custom Research regularly, and Kevin is already looking ahead to a larger customer validation initiative in 2027.

The plan: larger reports that show RingCentral's impact, ROI, and product benefits by segment across its small business customers, the core of its business. The aim is to show how small businesses in many industries use RingCentral to scale and compete with larger businesses.

Kevin credits G2's team as thought partners as well as execution partners, always watching for risks and gaps his team might not be thinking about. He also sees a quieter advantage in the AI interviewer itself.

"I felt that respondents were willing to be a little bit more honest with the AI interviewer because it's not going to take anything personally," says Kevin.

“You get the best of both worlds: quantitative results at scale from significant sample sizes, and qualitative depth from real verbatim responses. Together, they make up some of the strongest research available.”

Kevin Tsukii
Director of Content Marketing, RingCentral